Naloxone can effectively prevent death from opioid overdose if administered in time. However, less than 5% of people who need such treatment are receiving it as this life-saving treatment has to rely on someone near the unconscious victim to perform the administration. Therefore, mortality from an overdose is particularly high and the homeless or those living alone are disproportionately affected. To address this unmet need, many efforts have been made to develop automated naloxone delivery technologies. Unfortunately, due to the complexity of opioid toxidromes as well as the lack of multi-sensor modalities, these automated technologies tend to make extreme decisions for administering naloxone that are either too aggressive or too conservative. Aggressive decisions based on a false-positive detection of overdose would lead to unpleasant abrupt withdrawal syndrome. While conservative decisions made after extended monitoring often result in delayed administration of naloxone, which in turn increases the risk of brain hypoxia. Once a decision is made, a large dose of naloxone is necessary to enable rapid recovery from an overdose. In response, this work proposes a robotic subcutaneous implant composed of a variety of sensors to continuously monitor various cardiovascular and respiratory signals for real-time detection of overdose and a rapid drug delivery system for fast administration of naloxone. A decision-making algorithm for opioid overdose detection is developed based on the in vivo experiments of over thirty pigs overdosed by fentanyl. A barrel-cam like pump is developed to enable rapid delivery of 1 mL of naloxone within 5 sec. In addition, this work also proposes an algorithm that optimizes the energy consumption for continuously operating the sensors without compromising the detection accuracy to maximally prolong the battery life-time.